Microclustered optimistic simulation

Software-driven simulation as a design and verification tool has become almost ubiquitous in today's industrial and technology-based society. As systems grow in scale and complexity, often on a non-linear basis, demands on simulation change accordingly. Distributed simulation is one approach to deal with such growth.;Noting the limitations of static partitioning and contrasting dynamic re-partitioning, this thesis approaches simulation efficiency by proposing an initial over-partitioning of the problem. Using run-time data, we hope to discover potential efficiencies and exploit them by dynamically amalgamating these "micro-clusters" as suggested by run-time analysis.;As distributed simulation becomes the norm in various contexts, focus has shifted to the efficiencies of such simulations. A popular approach has been to partition a given problem into sub-units in such a manner as to increase the efficiency of the simulation run. Further, efforts have been made to dynamically re-partition the problem units using run-time metrics.